New York AI Prescribing Errors Up 15% in 2025

The integration of artificial intelligence into healthcare, particularly in medication management, promises efficiency but introduces novel risks. In New York, the rise of AI prescribing tools has coincided with a concerning uptick in reported errors, raising critical questions about medical malpractice. How prepared are our legal systems to address the complexities of AI-induced harm in patient care?

15% Increase in AI-Related Medication Errors Reported in New York State

A recent report from the New York State Department of Health (NYSDOH) revealed a 15% increase in medication error reports directly attributed to AI-assisted prescribing systems between 2024 and 2025. This figure, while not representing the entirety of all medication errors, points to a distinct and growing category of concern. The NYSDOH report, accessible via their official website, detailed instances ranging from incorrect dosages for common antibiotics to dangerous drug-drug interaction alerts missed by AI algorithms in complex patient profiles. My professional interpretation is that this isn’t merely a statistical anomaly. It signifies a systemic challenge in the deployment and oversight of these technologies. Healthcare providers, eager to adopt tools that promise to reduce administrative burden and enhance diagnostic accuracy, may be overlooking the critical need for strong human review and fail-safes. The sheer volume of prescriptions processed daily in major New York medical centers, like NewYork-Presbyterian Hospital or Mount Sinai Hospital, means even a small percentage of AI-generated errors can affect a significant number of patients. This data should serve as a stark warning: the promise of AI does not negate the physician’s ultimate responsibility.

Only 30% of New York Physicians Feel Adequately Trained in AI System Oversight

A survey conducted by the Medical Society of the State of New York (MSSNY) in early 2026 found that only 30% of practicing physicians in the state felt adequately trained to oversee and critically evaluate prescriptions generated by AI systems. This lack of confidence is alarming. Many practitioners, particularly those who completed their medical education before the widespread adoption of AI in clinical settings, report feeling ill-equipped to identify subtle errors or algorithmic biases. They understand the fundamental principles of pharmacology, certainly, but working through the black box of an AI’s decision-making process is a different skill entirely. This gap in training directly contributes to the potential for AI prescribing errors to go undetected. If a physician blindly trusts the output of an AI without understanding its limitations or how to cross-reference its recommendations with established clinical guidelines, they are effectively delegating patient care to an algorithm they don’t fully comprehend. This isn’t a condemnation of AI. It’s a call for mandatory, complete training programs focused on AI literacy and critical evaluation for all medical professionals using these tools. The legal implications are clear: a physician who fails to exercise due diligence in reviewing an AI-generated prescription, especially when presenting with clear red flags, could be found negligent.

Hurt by a medical mistake?

Know what your case is worth with AI Medical Payout Calculator for FREE!

Start my free evaluation

New York State Board of Medical Examiners Clarifies Physician Liability

In late 2025, the New York State Board of Medical Examiners issued a formal advisory reinforcing that the ultimate responsibility for a patient’s care, including medications prescribed, remains with the licensed physician, regardless of AI involvement. This guidance, published on the New York State Education Department’s Office of the Professions website, explicitly states that AI is a tool, and its recommendations must be critically evaluated by a human practitioner. This position is not surprising. It aligns with existing principles of medical malpractice law in New York, which generally hold the treating physician accountable for their decisions and actions. What this means in practice is that a physician cannot simply point to an AI system and claim it was the cause of an error. The expectation is that a competent physician would have identified and corrected the AI’s mistake. This creates a complex legal field. Proving medical malpractice in such cases will often involve demonstrating that the physician’s review process was substandard, or that they failed to act on clear warnings or discrepancies that a reasonably prudent physician would have recognized. It places a significant burden on practitioners to not only understand medicine but also to understand the limitations and potential pitfalls of the technology they employ.

Legislative Proposals Aim to Define AI Developer Liability by End of 2026

Recognizing the evolving legal challenges, several legislative proposals are currently under review in the New York State Assembly and Senate, aiming to clarify the liability of AI developers in cases of AI prescribing errors. One notable bill, introduced by Assemblymember Amy Paulin, seeks to establish a framework where manufacturers of AI prescribing software could be held partially liable if their product contains demonstrable defects or design flaws that directly lead to patient harm, provided the healthcare provider used the software as intended. This is a significant shift from the current model, which largely focuses on the end-user (the physician). The conventional wisdom has been that software is merely a tool, and the user is solely responsible for its application. However, I disagree with this narrow view when it comes to AI in healthcare. These are not simple word processors. They are complex systems designed to make critical medical recommendations. If an AI system is fundamentally flawed, perhaps due to inadequate testing, biased training data, or a lack of strong error-checking mechanisms, then the developer should share some responsibility. This isn’t about absolving physicians of their duties, but about ensuring accountability from all parties involved in the chain of care. Without holding developers accountable, there’s less incentive to invest in the rigorous safety protocols necessary for such powerful tools. While these proposals are still in committee, their passage by late 2026 could fundamentally alter how medical malpractice cases involving AI are litigated in New York.

The Conventional Wisdom: AI is Just a Tool, Physician is Always Liable

The prevailing sentiment in the legal and medical communities, as I often hear it discussed in continuing legal education seminars and medical ethics forums, is that AI is merely a sophisticated tool. Therefore, the physician who uses it bears full and sole responsibility for any errors. This perspective, while rooted in established legal precedent regarding medical instruments and diagnostic aids, fails to fully grasp the unique nature of AI. Unlike a stethoscope or an X-ray machine, AI doesn’t just provide data. It interprets, analyzes, and often generates specific recommendations, sometimes even drafting the prescription itself for physician approval. This is not a passive tool. It actively participates in the decision-making process. My disagreement with this conventional wisdom stems from the opacity and complexity of AI algorithms. A physician can understand the mechanics of an MRI machine, but few can truly “look inside” the neural network of a sophisticated AI prescribing system to fully comprehend its reasoning or potential biases. Holding physicians solely liable for errors originating from deeply embedded algorithmic flaws, which they have no means to detect or correct, seems fundamentally unfair and potentially counterproductive. It discourages innovation by placing an undue burden on practitioners, and it absolves developers of their responsibility to produce safe and reliable medical technology. We need a more nuanced legal framework that acknowledges the shared responsibility in the AI-human collaborative model of care. The notion that a physician can always catch an AI’s error presupposes perfect human oversight, which is an unrealistic expectation, particularly in high-volume clinical environments. Working through the complexities of AI-related medical malpractice in New York requires a deep understanding of evolving technology and established legal principles. If you or a loved one have been affected by a medication error potentially linked to AI prescribing, seeking immediate legal counsel is essential to understand your rights and the viability of a claim.